PONE-D-20-20631

People prefer simpler content when there are more choices: A time series analysis of lyrical complexity in six decades of American popular music

PLOS ONE

Dear Dr. Varnum,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. 

The two reviewers provide constructive and partially overlapping comments on your framing and the analyses. I strongly encourage you to consider the additional analyses and validity checks proposed by reviewer 1 as well as addressing the conceptual questions raised by both reviewers 1 and 2.

I am also wondering whether genre and the proliferation and diversification of genres over the last century may partially be responsible for some of these effects. To what extent do these trends occur within genres or over the careers of artists/groups? Do novel genres have an advantage over more established genres? Greater attention to genres of music as well as trends for the same agent (singer/songwriter, performer) may help to address some of the conceptual issues identified by the reviewers.

Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Nov 01 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

• A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Ronald Fischer

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

3. Please remove your figures from within your manuscript file, leaving only the individual TIFF/EPS image files, uploaded separately.  These will be automatically included in the reviewers’ PDF.

4. Please amend your list of authors on the manuscript to ensure that each author is linked to an affiliation. Authors’ affiliations should reflect the institution where the work was done (if authors moved subsequently, you can also list the new affiliation stating “current affiliation:….” as necessary).

Additional Editor Comments (if provided):

This is an innovative and thought provoking article. The two reviewers provide constructive and partially overlapping comments on your framing and the analyses. I strongly encourage you to consider the additional analyses and validity checks proposed by reviewer 1 as well as addressing the conceptual questions raised by both reviewers 1 and 2.

I am also wondering whether genre and the proliferation and diversification of genres over the last century may partially be responsible for some of these effects. To what extent do these trends occur within genres or over the careers of artists/groups? Do novel genres have an advantage over more established genres? Greater attention to genres of music as well as trends for the same agent (singer/songwriter, performer) may help to address some of the conceptual issues raised.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This study explores a trend towards greater compressibility of US song lyrics, which became more repetitive over the last 6 decades. The authors test the claim that this trend is due to an increase in the variety of songs on offer. The results show that novelty in music production (henceforth "musical novelty") is a significant predictor of lyrics compressibility, even when controlling, separately, for temporal autocorrelation on the one hand, and for a host of potential confounds on the other hand.

This is an exciting and innovating study, correctly done overall, and demonstrating an intringuing and non-trivial phenomenon: song lyrics become more repetitive over time. The use of a future-oriented predictive model is particularly appreciated. If the paper merely demonstrated and explored this trend I would have no reservations about it. My main concern comes from the causal hypothesis that the study puts forward to explain the trend.

The results only partially support the authors' claims. First, because the study fails to test a set of competing explanations that seem more plausible to me than the one put forward. They are detailed below. Second, because the claim that novel music production predicts lyric compressibility above other predictors (p. 18, "the amount of novel music produced contributes to changes in average lyrical compressibility above and beyond other plausible causes") is not demonstrated or even suggested by the data. Third, no evidence is given for the contention that more compressible songs are more likely to be successful, when there is more choice (in the authors' own data or elsewhere).

1. Alternative explanations

An explanation that is alluded to in one paragraph of the discussion (p. 21) but not followed through is that song lyrics became simpler and more repetitive because listening to music became something that people did while doing other things and often without paying any particular attention (in supermarkets, elevators, bars, etc., no longer just concert halls or standing on street corners). This would readily explain why lyrics become simpler: because songs no longer have the listeners' undivided attention. This explanation is entirely distinct from the hypothesised effect of musical novelty: it is about changes in music consumption, not about changes in music production. Even so, it is coherent with the pattern of results presented here. Arguably the musical industry produced increasingly many songs because demand grew, and demand grew because people took to listening to music in circumstances where they did not use to. Changes in media of diffusion (e.g. from sheet music to radio) are an obvious and related explanation. Unless we assume that these two hypotheses are somehow equivalent or interchangeable, one cannot claim that growing musical novelty caused the observed trend without ruling out this alternative account.

One may also worry about a possible selection bias. As explained in the supplementary materials, the study selected roughly half the songs that appeared in the charts for textual analysis, due to difficulties in finding good textual data for other songs. This raises the possibility that a selection bias might explain the observed trend. It is possible that text data is better for later songs: that our documentation for 2000s hits is better than it is for 1960s hits. It is possible that songs with more less compressible lyrics are more likely to be documented, because they are more interesting, lyrics-wise, and more worthy of attention. If these two conditions obtained they would suffice to produce an apparent decrease in compressibility that would be entirely due to a preservation bias. Lyric compressibility would not actually decrease through time for unrecorded song lyrics. I am not saying that this is what happened, but this explanation is easy to rule out (just show that the proprotion of hit songs with undocumented lyrics does not change through time, or that such changes, if they occur, do not explain away the trend you observe). Relatedly, more detail on the selection of song lyrics to be analysed would be welcome: what the criteria for inclusion were, whether there was any stopping rule for data collection, etc.

2. Is novel music production a better predictor of lyric compressibility than other predictors?

The results do not establish that musical novelty is a better predictor of lyrics compressibility compared to other possible predictors studied here. Several indicators show a higher correlation with lyrics compressibility, among them (judging by Fig. 1) GDP per capita, population size, and (with an inverse correlation) residential mobility. (Although I don't know what would happen to these correlations after autocorrelation is taken into account.) To sustain the claim that musical novelty is a better predictor of lyric compressibility than other candidates, running partial correlations is not sufficient. Partial correlations merely show that the correlation between lyrics compressibility and musical novelty is robust when variable X is taken into account, but it could still be the case that variable X does better, as a predictor of lyrics compressibility, than musical novelty does.

Relatedly, it is not clear whether the correlation between lyrics complexity and musical novelty would still hold once all important confounds are controlled for *together*, and not just separately as done here. The choice of analysis that was made for this study (taking years as data points) does not allow this to be shown (too few data points), but a nested regression taking songs as data points instead of years might allow the authors to demonstrate this (with due attention being paid to multicollinearity). Alternatively, the authors could reduce all the potential confounds (all factors listed in Fig. 1 except Lyric compressibility, Music production, and Year) to one super-factor, with a PCA. Showing that the correlation between lyrics complexity and musical novelty holds when doing a partial correlation controlling for this super-factor would help make the authors' point.

3. Missing evidence of greater success for simpler songs

On p. 3–4, the study justifies the hypothesis to be tested on the grounds that people generally prefer simpler content to more complex content, especially when the choice is broad. This debatable claim is made by analogy with results in social psychology and experimental economics which in my view are not clearly relevant to the material being studied here. The similarity between a simple economic decision (e.g. a financial product that is easy to understand, as in Iyengar & Kamenica 2010) and a repetitive song, seems quite remote to me. Still, this view makes one clear prediction: more compressible songs should be more commercially successful than compressible ones, at least when there is a lot of choice. The paper seems to endorse this point but does not cite any evidence for it. It would be easy to answer this question, by comparing billboard hit songs with non-hits and controlling for various other factors.

Minor comments:

One possible confound that is (in my view) unlikely to explain the study's correlations but is easy to control for and should be ruled out, is song length: given the measurement of compressibility, I suspect song length will strongly impact compressibility, and if there is any trend in time towards shorter or longer song this might confound the observed trends.

The legend for figure 1 says that the correlations between variables are given as Kendall's tau, but I doubt it for two reasons. 1: The value given in the figure for the correlation between the Music Production index and Lyric Compressibility is .88, which does not correspond to the value reported in the main text (Kendall’s τ = .714), but does correspond to the Pearson's r correlation given in the markdown file (Pearson's r = .87723). 2. In the source code for the figure the method for the correlation is not specified (the command is cor(years, use="" ext-link-type="uri" xlink:type="simple">pairwise.complete.obs")). I suspect R defaults to method = "pearson" when method isn't specified. Please clarify and correct if needed.

Correlations are occasionally (exceptionally) given using Pearson's r (p. 10, also p. 14 when reporting the results for Tiokhin-Hruschka method). The authors note that this parametric correlation is inappropriate since time-series data are not normally distributed. Please remove mentions of Pearson's r or uses of it in reporting results. I recommend paying special attention to results on the Tiokhin-Hruschka method when doing so. See also the above comment regarding Fig. 1.

p. 16 AIC stands for Akaike's Information criterion (not Aikeke).

p. 20 This passage of the discussion alludes to a section of the supplementary materials that I could not find: "the aim of the present work was to understand what shapes the success of cultural products over time, rather than to use the broadest possible set of cultural products as a way to gain insight into other phenomena at the population level (see supplement for an extended discussion of this issue)."

Reviewer #2: This paper presents an analysis of why pop music in the US has become lyrically simpler over time, testing the hypothesis that the trend is driven by an expansion in the number of available song choices. This is tested by quantifying lyrical simplicity using a metric of information compressibility (LZ77 compression algorithm) over thousands of songs, and correlating this measure with estimates of the number of new songs in each year. The results support the hypothesis: large correlations between the measures.

The paper is well written and the analyses are sound and generally appropriately interpreted. The ‘multiverse’-style analysis approach is also helpful in that it provides converging different approaches. The results will be of interest to people in the psychology of music, cultural evolution, and the general public as well.

Here are a few suggestions for a revision:

(1) What songs are most popular and make it to Billboard is not unrelated to preferences, but also not that tight of a measure of people’s self directed-listening behaviours and preference for music, as is implied by the use of "preferences" throughout the paper. for instance, radio plays are influenced by advertisers, independently of people's preferences for songs. A tighter claim to make is that, as more music becomes available, simpler songs are more memorable and/or dispersible than more complicated ones. Whether and how this is related to claims in the manuscript about peoples’ music preferences changing based on Kahneman-esque heuristics being deployed due to increased cognitive load (Intro, pages 4 and 5) and/or interpreting these changes in lyrical trends as indicating changes in emotional expression (if this is what the abstract framing + discussion is implying? Eg. in “What does this tell us more broadly about how American culture has changed?”) is more up for debate, I think. This is an easy fix: just need to clarify the interpretation in the paper a bit more.

(2) The manuscript is clear that the correlational data doesn’t justify claims about causality, but it would be helpful to tighten up the areas where an interpretative claim is being made. Might the direction of causality be backwards? Songs that are simple could be easier to produce, so as artists realize they can produce simpler styles, maybe they produce more of them? There are plenty of other explanations here that would be good to discuss. For instance, maybe memorability is a big driver in what songs get a lot of radio plays, where memorability is a different aspect of music perception than preference.

(3) There may be some interesting parallels to be drawn between these results and ongoing research in how languages more generally are shaped by communicative efficiency (see for review: Gibson et al., 2019, TICS). Namely, the primary measure of simplicity of lyrics is sensitive to word length. Zipf’s law describes the frequency structure of words in a language as being related to word length (eg, Piantadosi, 2014, Psychonomic Bulletin Review), although more recent work shows that information content of words is a better predictor of word length than frequency-rank (Piantadosi et al., 2011, PNAS): in other words, more predictable words tend to be shorter. Something like Zipf's law is at work in music (see Levitin et al., 2012, PNAS; Mehr et al., 2019, Science) and so this connection with information-theoretic notions of communication would be productive. (It also fits neatly with how lyrical simplicity is quantified with LZ77).

(4) To what extent is variance in lyrical compressibility in these data mediated by the distribution of genres within the presented dataset? Electronic/dance music often has highly simple repetitive lyrics as a defining feature, for example, more so than, e.g., jazz lyrics. Perhaps one of the reasons for the popularity of electronic/dance genres within the broader popular music space may relate to this claimed attraction toward simplicity of lyrics. But the deeper point is then to ask how much of the variance in lyrical compressibility is stemming from a general trend across popular music genres and how much is contributed by relative shifts in other stylistic factors (that may be correlated with greater lyrical compressibility for additional reasons). Disentangling this is probably difficult, but I feel like it could be discussed.

Minor comments:

For the predictions about the lyrical compressibility of future popular music, some comments about the bounds in which such extrapolation is valid/meaningful would be helpful. What does it mean for music to have an average compressibility index of ~1.225 by 2050 (as compared to the current average of ~1.1)? What are reasonable bounds of compressibility that things might plateau at?

Please check references, as at least one in-text citation was not in the end references (Steegen et al., 2016)

Mehr Krasnow 2017 is a bit of a funny citation for "music is a human universal". I think better might be Mehr et al., 2019, Science and/or the new BBS theoretical treatment (https://doi.org/10.1017/S0140525X20000345)

A reference about how lyrics play an important part in people’s listening habits may be helpful. For instance, this paper based on Spotify listening data would be a helpful citation: http://archives.ismir.net/ismir2018/paper/000098.pdf.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Why are song lyrics becoming simpler? A time series analysis of lyrical complexity in six decades of American popular music

PONE-D-20-20631R1

Dear Dr. Varnum,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Ronald Fischer

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Congratulations, I recommend your article for publication to the Editor in Chief.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: All my comments were addressed more than satisfactorily. The authors are to be congratulated for this excellent contribution!

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Olivier Morin

Reviewer #2: No